��֭����

This report explains ��֭����’s survey — covering over 200 public universities and colleges — concerning how these government institutions filter or block content and users on their official Facebook and Twitter accounts. Using examples of censorship identified in public records acquired from these institutions by ��֭����, “No Comment: Public Universities’ Social Media Use and the First Amendment” discusses how the First Amendment applies to government actors’ use of social media and explains why their use of the filtering and blocking tools may violate the First Amendment.

Executive Summary

In October 2019, Facebook founder and chief executive officer Mark Zuckerberg spoke at Georgetown University, extolling the virtues of freedom of expression and noting in particular the importance of college students’ ability to “express who they were and what mattered to them,” including through “challeng[ing] some established ways of doing things on campus.”^[1]

Because Facebook is a private entity, the First Amendment—which only limits government actors—does not require it to honor expressive freedom. Zuckerberg’s endorsement of freedom of expression as a principle is a welcome and encouraging development.

It is, however, at odds with the fact that Facebook provides governments the tools to censor. These actors include public universities and colleges which are bound by the First Amendment—those very campuses where students have “challenged some established ways of doing things.”

These tools include Facebook’s automated content filters, which allow state institutions to automatically “hide” users’ comments if they contain words included on Facebook’s undisclosed list of offensive words or the government actor’s customized list of prohibited words. These tools enable public universities—and other government actors—to quietly remove critical posts, transforming the Facebook pages into less of a forum and more of a vehicle for positive publicity.

This censorship has deleterious effects on campus and public discourse. Using public records requests, FIREsurveyed over 200 public universities and colleges across 47 states and the District of Columbia. Half of the surveyed institutions choose to use Facebook’s “strong” profanity filter, and nearly a third—55, or 27.8%—use the “medium” filter. This means that 77.4% of surveyed institutions choose to employ a blacklist of prohibited words not disclosed to the public.

Additionally, nearly a third of the universities surveyed (59, or 30.3%) use a custom blacklist, collectively censoring 1,825 unique words and phrases. These range from familiar profanities to words relating to matters of local and national concern—for example, blocking animal rights activists’ criticism of food vendors, curbing discussion of what to do about a campus Confederate monument, and shielding criticism of itinerant preachers, controversial faculty, politicians, and sports teams.

Even without these tools, public actors can prune their Facebook comments, manually hiding comments to manipulate and shape the apparent public discourse. Wright State University, for example, deleted comments supporting a faculty strike from its Facebook page, leaving behind a community “forum” that largely (and falsely) appeared supportive of the university’s administration and critical of striking faculty.

Because Facebook doesn’t alert a user when their post has been removed, or tell the public that comments have been censored, these commenters (and other users) may never know their words have vanished.

Further, both Facebook and Twitter allow government actors to block members of the public from participating in these public forums.

These automated methods of censorship are not only contrary to a commitment to freedom of expression, but also provide government actors with tools that—in light of recent federal court rulings concerning President Trump’s Twitter feed—violate the government actors’ legal obligations under the First Amendment.

Methodology

FIRE issued public records requests to five institutions in each state where public institutions of higher education are subject to public records laws, which are similar to the commonly known federal Freedom of Information Act (FOIA).

The five institutions per state were divided into two categories:

1. The three public, four-year institutions with the largest undergraduate enrollment; and
2. The two public, two-year institutions with the largest undergraduate enrollment.

Our approach has some logistical limitations. Some states, like Delaware and Pennsylvania, exempt some or all of their institutions of higher education from their public records laws, or limit the scope of the laws to specific expenditures. Other states’ public records laws expressly limit their public records laws to requests made by citizens of the state.^[2] In other states, there are fewer than five institutions, or fewer than the three four-year and two two-year institutions this survey sought to cover. Finally, some institutions could not be contacted due to the potential for conflict with a pending FIRElawsuit against the institution. In those cases, the institution was replaced with the public institution with the next-highest enrollment. A full list of exceptions may be found in Appendix B.

Altogether, ��֭����’s survey issued public records requests to 224 public universities and colleges in 47 states and the District of Columbia. 198 institutions provided substantive responses. A full list of surveyed institutions may be found in Appendix A, which also identifies the institutions which failed or refused to produce records and includes data compiled from the survey.

Our public records requests identified the official Facebook and Twitter account for the institution and asked the institution to provide to FIREthe Facebook settings, list of blocked Facebook users, and list of blocked Twitter users. An example of a representative request may be found in Appendix C. We encourage journalists or interested members of the public to adapt our request to ask other public institutions—not just universities—about their records.

The calculation of the number of words on institutions’ customized blacklists excludes duplicates. Thus, if two institutions blocked the word “dog,” that word is counted only once. However, the calculation of the number of blocked Facebook users, as well as the number of blocked Twitter users, did not exclude duplicates. Accordingly, if one user were blocked by two universities, she would be counted twice.

Discussion

Facebook's tools

Facebook allows institutions, entities, organizations, and government actors to establish pages, operating as a forum where the public can share their thoughts with or about the organization. This includes public officials and government entities, ranging from the President of the United States to the City of North Pole, Alaska. As described below, public actors are bound by the First Amendment when they use tools provided by private entities.

When a business, organization, government entity, or public figure creates a “page,” Facebook allows them to create posts on their own page, gives the option of allowing others to add their own posts, and allows Facebook users to add comments to those posts.

Facebook provides page owners with four tools to limit user content:

1. The profanity filter. Facebook’s “profanity filter” automatically hides ���������ٴǰ���’ posts if they contain words on one of two lists—one for the “medium” setting and one for the “strong” setting.^[3] The words on these lists are not publicly disclosed, but are composed of “the most commonly reported words and phrases marked offensive” by Facebook users.^[4] The profanity filter is turned off by default.
2. The customized blacklist. The “page moderation” filter allows an administrator to establish a custom list of blocked words.^[5] Like the profanity filter, this filter automatically hides posts or comments if they contain a phrase on the custom list.
3. The blocking function. An administrator of the page may ban particular accounts, after which “they'll no longer be able to publish to [the page], like or comment on your . . . posts, message [you] or like your [page].”^[6]
4. Manual removal. An administrator may manually hide specific posts or comments.

A user may not know that their comment has been hidden, as it remains visible to the poster.

Twitter's tools

Twitter offers fewer methods of regulating user interactions with government accounts:

1. The blocking function. Twitter users can block others from interacting with them.^[7]
2. The muting function. Twitter users can mute users and terms, but this function does not prevent those users from interacting with the account, and a muted user’s tweets can still be seen by other visitors.^[8]

The First Amendment limits public actors’ use of filtering tools

With the emergence of the internet and social media, courts have increasingly been called upon to apply the First Amendment to the digital realm. As the Supreme Court of the United States recently observed, “in the past there may have been difficulty in identifying the most important places (in a spatial sense) for the exchange of views,” but the answer today is “clear”: “It is cyberspace . . . and social media in particular.”^[9]

Accordingly, a wide range of courts across the country have held government actors’ social media sites,^[10] including those on Facebook and Twitter, to be subject to First Amendment limitations.^[11] The most prominent of these decisions thus far is a successful challenge to President Trump’s practice of blocking critics from his Twitter account.^[12] These rulings only limit government actors’ use of social media tools; the First Amendment does not impose legal obligations on social media sites like Facebook and Twitter, nor on other private parties when they use social media.^[13]

What is the public forum doctrine?

The general thrust of the cases addressing public actors’ use of social media is that the “interactive space” constitutes a “public forum.” This is a term of art in First Amendment law, setting forth a “metaphor . . . first used in constitutional free speech cases as a way of explaining why the government cannot engage in . . . content discrimination with regard to speaking, picketing, or leafleting on city parks and sidewalks.”^[14]

A public forum is most often seen as a physical site. For example, a public sidewalk or park would likely be seen as a “traditional public forum”: places where members of the public may freely gather and talk about whatever is on their minds.^[15] In those places, attempts to limit viewpoint are never permissible, and attempts to limit content must meet “strict scrutiny”—that is, any limits must be necessary to address a compelling government interest and narrowly tailored to serve that interest.

Meanwhile, a meeting space set up by a government entity for discussion of particular subjects or use by particular groups might be a “limited public forum.” For example, school board meetings at which members of the public can speak are limited public forums, as they’re set aside for a particular purpose.^[16] In these spaces, government has a freer hand to regulate the subject matter of speech or who may utilize the space, but can’t exclude people or speech based on the viewpoint expressed. Even if a particular space is privately owned, it is a public forum when a government actor exercises control over it, such as when a city leases a private theater.^[17]

Public forum doctrine distinguishes between speech by the government and its employees—“government speech”—and speech by others that the government regulates.^[18]

What are the different standards for different public forums?

Not every space opened to expression is a free-for-all in which anything goes. Different standards attach to the forum depending on its purpose, and the extent to which government actors may regulate expression within that forum depends on the forum’s purpose and the expression at issue:

• Government actors can almost always limit expression that falls into one of the historic, narrowly-defined exceptions to the First Amendment: “incitement, obscenity, defamation, speech integral to criminal conduct, so-called ‘fighting words,’ child pornography, fraud, [and] true threats.”^[19] This is true in all forums: traditional, designated, and limited.
• Government actors can never limit expression because of its “point of view”^[20] or where the “specific motivating ideology or the opinion or perspective of the speaker is the rationale for the restriction.”^[21] This is also true in all forums.
• If no effort is made to set aside the space for discussion by particular people or about particular subjects, the space is likely to be treated as a designated or “open” public forum.^[22] In that case, restrictions on the “time, place, and manner of expression” are permissible only if they are “content-neutral, are narrowly tailored to serve a significant government interest, and leave open ample alternative channels of communication.”^[23] If a burden on speech is content-based, the regulation is permissible only if it is “necessary to serve a compelling state interest and that it is narrowly drawn to achieve that end.”^[24]
• If a policy is established and consistently enforced^[25] to limit use of the forum for particular people or discussion of certain subjects, the space is more likely to be treated as a limited public forum, in which the government has greater leeway to regulate speech.^[26] There, restrictions need not be content-neutral, but must be viewpoint-neutral, “reasonable in light of the purpose served by the forum,”^[27] and must “comport with the definition of the forum”^[28]—that is, the government cannot exclude speech for which, or speakers for whom, the forum was opened.

These standards are also important for evaluating whether the regulation of online expression is permissible under the First Amendment.

How does the public forum doctrine apply to social media?

While public forums are most easily conceptualized as physical spaces—such as open areas, meeting rooms, and bulletin boards^[29]—the public forum doctrine has also been applied by the Supreme Court to forums which consist “more in a metaphysical than in a spatial or geographic sense.”^[30] For example, the same principles have been applied to—albeit with varying outcomes—student activity fees pooled for student organizations,^[31] schools’ internal mailing systems,^[32] and email systems.^[33] As governments have ventured into cyberspace, courts have applied these principles to government websites, distinguishing government-published websites from websites that allow members of the public to exchange views.^[34]

Courts have used the legal framework provided by public forum doctrine to analyze government social media use. While the government actor has control over its own posts (which are quintessential “government speech”), the “interactive space” that follows those posts and allows for response from the public constitutes a public forum subject to the First Amendment’s protection.^[35] In other words, while a public university student or faculty member might have a right to post a relevant comment on an existing thread, they could not compel the university to “share” their post in the same manner as the university publishes its own posts.

Here’s a visual demonstration of how “government speech” and the interactive digital spaces work on Facebook (left) and Twitter (right):

Because the First Amendment protects users’ speech in these interactive spaces, government actors like public colleges and universities cannot censor user comments in that space because they disagree with the viewpoint expressed. That means that—at a minimum—government actors cannot block users,^[36] remove posts,^[37] or otherwise “burden” speech^[38] because they find the expression offensive, disagreeable, or wrong.

That does not mean that every digital space associated with a government actor is a free-for-all forum where any content may be shared. How government actors characterize and treat digital spaces is important, as regulations on interactive spaces will be analyzed based on whether they are—as with physical spaces—traditional, dedicated, or limited public forums.

If the interactive space is a limited public forum, some restrictions will be permissible. For example, a prohibition on “clearly off topic” comments is a “self-evidently viewpoint-neutral” regulation “limiting a forum to discussion of selected topics,” and is “reasonably related — indeed, integral — to the forum’s purpose.”^[39] As such, these kinds of viewpoint-neutral limitations are constitutionally permissible and do not violate the First Amendment.

But the government’s ability to regulate speech in a limited public forum has boundaries. If, in crafting a policy governing social media content, the policy affords too much discretion to the institution’s content moderators, those standards will inevitably be abused to censor criticism of the institution. The University of Connecticut’s policy, for example, reserves to officials “” to remove comments they believe to be “,” expression that “,” or posts that are “.”

These types of broad, ambiguous policies will lead to viewpoint discrimination and may render the policy itself unconstitutional.^[40] For example, Wright State University created “guidelines” barring comments deemed to be “propaganda,” “foul,” “trolling,” “offensive,” or “inflammatory,” giving staff members—overseen by university administrators—unfettered discretion to manually hide comments criticizing the administration’s positions during a faculty strike, such as those disclosing senior administrators’ salaries.^[41] (At the time of publication, Wright State officials had communicated to FIREthat the university was conducting an internal review of its Facebook guidelines in response to a letter from FIREraising concerns about the policy.)

In addition to the ability to limit expression by policy and manual enforcement, Facebook provides a variety of automated content-policing tools to administrators of Facebook pages.

However, the tools provided by Facebook are not adequately tailored and do not respect the contours of the First Amendment, and their use by government actors will almost certainly lead them to violate the First Amendment rights of students, faculty, and the general public. When used by public institutions, Facebook’s tools amount to always-on sentinels tasked with identifying and automatically censoring student and faculty speech—and, for years, government actors have been using these tools to effectuate online censorship.

Public institutions are using Facebook and Twitter tools to violate the First Amendment

It has long been settled law that the First Amendment is binding on public colleges and universities, including when they act to regulate the otherwise-protected speech of students, faculty, and organizations comprised of students or faculty members, such as student clubs or faculty unions.^[42] When public universities open digital spaces for discussion by faculty, students, and the general public, regulations of those digital spaces must meet First Amendment scrutiny.

Beginning in October 2018, FIREissued requests under public records laws to 224 public universities and colleges in 47 states, plus the District of Columbia. The selected institutions represent the three four-year institutions and the two two-year institutions with the highest enrollment in the state.^[43]

Of the 224 institutions surveyed, 198 (or 88.4%) provided substantive responses.

FIRE found that many public universities and colleges are using automated tools to regulate online expression in ways that do not comport with the First Amendment.

Most public colleges use Facebook’s ‘profanity’ filters, secret blacklists of words that risk violating the First Amendment.

Half of the responding institutions—99 of 198—use Facebook’s “strong” profanity filter, and nearly a third—55, or 27.8%—use the platform’s medium filter. This means that 77.8% of surveyed institutions use some version of the profanity filter.

Because Facebook does not disclose the list of words or phrases on either the “medium” or “strong” profanity filters, it is not clear what content is automatically scrubbed from the pages of institutions that utilize the filters created by Facebook. (The custom filters, described below, can be identified through public records requests.)

Facebook, again, does not violate the First Amendment when it creates these filters and offers them to their users. Instead, the First Amendment protects the rights of private entities like Facebook to decide what speech to publish or refuse.^[44]

However, when a government agency uses these tools to filter particular words or phrases, it raises serious First Amendment concerns. The premise underlying this type of restriction—that blocking individual words is justifiable under the First Amendment because it will improve public discourse—has been expressly rejected by the Supreme Court. In Cohen v. California, the Court held that a jacket emblazoned with the words “Fuck the Draft,” worn in a courthouse hallway populated with “women and children,” was protected speech.^[45] “[W]e cannot indulge the facile assumption that one can forbid particular words without also running a substantial risk of suppressing ideas in the process,” the Court explained, as “governments might soon seize upon the censorship of particular words as a convenient guise for banning the expression of unpopular views.”^[46]

This risk is most colorfully illustrated by institutions’ use of customized blacklists to suppress discussion of local controversies, as discussed below. However, the broader use of Facebook’s own profanity filters—a secret list of disapproved words automatically compiled from Facebook users’ feedback—will have the same effect, while failing to meaningfully protect equitable, civil discourse.^[47]

First, because the list is generated by compiling the terms “most commonly reported” as offensive, it necessarily limits speech that is in the greatest need of protection from censorship because it is disapproved by those whose views and sensibilities are more likely to be majoritarian or mainstream. As the Supreme Court observed in holding that burning the American flag is protected expression, “[i]f there is a bedrock principle underlying the First Amendment, it is that the government may not prohibit the expression of an idea simply because society finds the idea itself offensive or disagreeable.”

The use of these databases may also, incidentally, inure to the detriment of minority speakers. One recent study, for example, found that “tweets by African American authors are 1.5 times more likely to be labeled ‘offensive,’” and tweets in African American English are “more than twice as likely to be labeled as ‘offensive’ or ‘abusive.’”^[48] As a result, automated means of identifying human speech, when relying on human-generated data, may introduce “human biases [which] can easily result in a skewed distribution in the training data,” yielding “unintended bias in the resulting models, and therefore potentially unfair applications.”^[49] Because datasets of offensive language may yield a “systemic racial bias,” they may “have a disproportionate negative impact” on minority communities, effectively creating systems that “may discriminate against the groups who are often the target of the abuse we are trying to detect.”^[50]

Second, an automated system is unlikely to interpret the context and meaning of uses of words that may be offensive in some contexts and inoffensive in others. Content “moderation is, inherently, a subjective practice,” rendering “content moderation at scale . . . impossible to do well.”^[51]

As a journalist reporting on recent studies on attempts to create filtering mechanisms for offensive speech observed:

[W]hat is considered offensive depends on social context. Terms that are slurs when used in some settings — like the “n-word” or “queer” — may not be in others. But algorithms — and content moderators who grade the test data that teaches these algorithms how to do their job — don’t usually know the context of the comments they’re reviewing.^[52]

Similarly, automated systems are incapable of making value judgments, resulting in the suppression of speech that makes use of offensive words in order to appropriate or report on that speech. While a filter may be intended to prevent, for example, the use of a racial slur directed to another user, it will also suppress posts by a student who, targeted by use of that same slur, posts about their experience on their university’s Facebook page to call on the university to take action. Activists utilizing offensive language or imagery in order to report on or criticize their use may not anticipate automated filtering, but a user dedicated to using the language or imagery in order to harass others may be more likely to devise workarounds, or use language less likely to be caught in a filter.

Third, the use of blacklists almost certainly fails First Amendment scrutiny under either a designated or limited public forum analysis. Facebook’s profanity filter is expressly premised on feedback identifying which words or phrases users find offensive. A limitation on this basis is not viewpoint-neutral, rendering its use fatal in any forum.^[53] Further, it is unlikely that a public institution will be able to bear its burden—to demonstrate that its restriction is reasonable—if it cannot identify the words that it is restricting. Similarly, an undisclosed list of prohibited terms necessarily means that members of the public do not have notice as to what speech is permitted or prohibited, ensuring that the restriction will “trap the innocent by not providing fair warning” about what is prohibited.^[54]

An automated blacklist also fails First Amendment scrutiny because it automatically hides speech unless it is reviewed by an administrator of the page, effectuating a system of prior restraints on speech, “the most serious and the least tolerable infringement on” freedom of expression.^[55] The risk prior restraints present to freedom of expression is so great that the “chief purpose” in adopting the First Amendment was to prevent their use.^[56] Prior restraints are permissible only where they are bound by “narrow, objective, and definite standards”^[57] and where the authorities’ review and determination must be made within a specific period of time.^[58] Absent these procedural safeguards, which FIREhas not observed at any surveyed institution, a prior restraint is unconstitutional.

An automated system that bars specific words does not differentiate between offensive uses of a word or uses of profound public importance. While it may serve a college or university’s public relations goals to police public-facing expression by students and faculty for civility, it undermines the institution’s commitment to freedom of expression and obligations under the First Amendment.

Facebook’s customizable blacklists are used to restrain speech critical of government institutions and their corporate partners, or speech on matters of public concern.

Nearly a third of the public colleges and universities FIREsurveyed (60, or 30.3%) use a customized blacklist on their Facebook pages, collectively censoring 1,825 unique words and phrases. These range from the words populating George Carlin’s famous list of “Filthy Words”^[59] to words and phrases pertaining to political and social matters of local and national concern. Together, these customized lists evidence use of Facebook’s technology to automatically censor criticism of institutions, corporate partners, campus controversies, and even sports-related chest-thumping.

These filters are not likely to pass First Amendment scrutiny. Moreover, many of these filters evidence viewpoint discrimination, as they are specifically designed to hide terms closely associated with particular criticisms. For example, at the University of Kentucky are inarguably intended to suppress criticism by animal rights activists, even if a supporter of the university’s vendor would also face restrictions because of a post containing the same term.

Even where a filter is not likely to amount to viewpoint discrimination, its use is not likely to be sufficiently tailored to meet the institution’s obligations under the First Amendment.

For example, a student at the University of North Carolina at Chapel Hill might want to respond to a university post about Black History Month^[60] to raise questions about “Silent Sam,” a Confederate monument that until recently stood on UNC’s campus. But they could not have done so if their comment used the monument’s name, because that phrase was on UNC’s Facebook blacklist.^[61] Or, similarly, when the university posted a message announcing the removal of the monument’s pedestal,^[62] comments containing the words “Silent Sam” would unquestionably be relevant. Yet none of the hundreds of visible comments contain those words—almost certainly because of the blacklist. Similarly, students who used the phrases “9/11” or “terrorism” in responding to the university’s post memorializing the September 11 attacks^[63] would not reach their intended audience—again, almost certainly because of the blacklist. Yet each of these comments would be on-topic and not otherwise fall within a categorical exception to the First Amendment.

Similarly, students at schools like Oklahoma State University, where the names of political candidates are automatically scrubbed, would encounter difficulty if they mentioned their favored (or disfavored) candidates’ names in response to the schools’ posts about getting out the vote.^[64]

These restrictions, even where they are viewpoint neutral, are not “reasonable in light of the purpose served by the forum”^[65] because they burden relevant, on-topic expression in a medium intended to operate in real-time. Instead, they effectuate a system of prior review on particular terms, and institutions are not likely to provide the procedural protections required by the First Amendment.

FIRE’s survey revealed that public institutions are using the customized blacklist feature to limit a wide range of criticism, or to blunt discussion of local controversies:

• The University of Kentucky blocks the words “,” “,” “,” and “.” The university explained to FIREthat it instituted this restriction “around the time that Aramark came on campus” and activists began posting “highly graphic videos of chicken slaughter.” In 2014, the university announced it was entering into a 15-year contract worth $250 million with Aramark.^[66] Similarly, Mississippi State University blocks mentions of “” and .
• Texas A&M University blocked terms in an effort to bar criticism by animal rights activists, including People for Ethical Treatment of Animals, over research conducted on dogs. These terms include “,” “,” “,” and “.” PETA and the Electronic Frontier Foundation sued Texas A&M in 2018 over the censorship.^[67] Santa Monica College likewise bars “,” “,” “,” and “” following a PETA campaign critical of cat dissection in an anatomy course.^[68]
• During protests over the “Silent Sam” Confederate monument, the University of North Carolina at Chapel Hill blocked posts containing the phrase “,” as well as mentions of “.” The university also blocked mentions of the terms “,” “,” “,” and the name of a professor whose teaching about the September 11, 2001, terrorist attacks drew national media coverage when a student, who had not read any of the materials,^[69] publicly complained about the materials read in the course.^[70] The university also barred .
• The University of Arizona automatically removes posts or the name of an itinerant preacher known for holding signs reading “You Deserve Rape,” which has drawn media criticism.^[71] This restriction presumably removes complaints about the preacher.
• Suffolk County Community College (NY) blocks posts concerning inclement weather closures, and includes terms that demonstrate that the purpose is to inhibit criticism, barring the words “,” “,” “,” “,” “,” “,” “,” and “.” They also block “,” “,” and “.”
• Clemson University , a meme about a gorilla that had previously generated controversy on campus when a residential employee sought to bar students from referencing the meme on their dormitory doors. Clemson also blocked phrases referring to other controversies on campus, including the name of a professor criticized for calling Republicans “,” the word “” (apparently because of a student’s poll of his classmates, which was reported in the press, about purposefully driving over turtles). Clemson also blocks the words “” and “”—a company with which the university has entered into a $58 million contract.^[72] Similarly, the university blocked the words “,” “,” and “,” likely relating to students who protested by sitting during the pledge of allegiance during a student government meeting.^[73] Clemson also blocked the word “,” referring to who called on the university to end its practice of releasing balloons during football games.^[74] Ominously, Clemson also blocked the words “” and “.” In response to ��֭����’s records request, Clemson removed the blocked terms and implemented “measures so that blocks of protected speech do not occur” in the future.^[75]
• The once-popular awareness campaign “”^[76] cannot be found on Auburn University’s Facebook page, because comments containing that word are automatically deleted.
• The University of Mississippi prohibits mentions of its former mascot, “,” a “caricature of an antebellum Southern plantation owner” no longer used by the university.^[77]
• The University of Washington blocks a pejorative pun on its team name, the Huskies (“”), as well as references to its football rivals: “,” “,” “,” and “.” Texas A&M University blocks references to the “” horn gesture of their University of Texas at Austin rivals, the Longhorns. Oklahoma State University blocks mentions of its rival football team, including the phrases “” (OU’s fight song), “,” and “.”
• Not to be outdone by its rival, the University of Oklahoma an emoji: ?.^[78]
• The University of Central Florida blocks the words “” and “.”
• Oakland Community College in Michigan blocks mentions of the name of a disgruntled former professor.
• Mississippi State University blocks the term “,” a play on its “Hail State” motto.^[79]
• A number of institutions, including , , the , the , blocked the names of political candidates, such as “Trump,” “Bernie,” or “Hillary.” Others, such as , restrict pejorative versions of their names or supporters, such as “Trumptards” and “NObama.”
• West Virginia University blocks mentions of “,” a tradition long disfavored by the university’s administration and Morgantown residents.^[80]

Not just words: Public universities block thousands of Facebook and Twitter users.

FIRE’s records requests reveal that of the 198 public colleges and universities that responded to ��֭����’s survey, 173 (or 87%) blocked users on Facebook or Twitter. Together, these institutions blocked 13,197 Facebook users and 4,065 Twitter users from interacting with their posts, pages, or tweets. Accordingly, an institution that blocks at least one user blocks an average of 76 Facebook users and 23.5 Twitter users.

Because Facebook and Twitter do not keep records of the reason for blocking a particular account, and because institutions’ records reveal only the name of the blocked user (not what they posted), it’s unclear how many of these users were blocked for viewpoint-discriminatory reasons. It’s possible many were blocked for violating constitutionally-defensible regulations.

However, government actors cannot restrict access to a public forum based on an objection to the viewpoint of the speaker. Many blocked accounts relate to political, social, or religious movements, including many critical of the institution, raising the possibility that they were blocked for advocating a particular viewpoint.

For example:

• The University of Kansas blocks “Boycott Koch Industries,” a Facebook account referencing America’s second-largest privately-held company,^[81] which is based in Kansas and is a major donor to the university.^[82] On Twitter, the university blocks accounts related to the Westboro Baptist Church, which had the university’s graduation.
• Georgia State University blocked a “” Twitter account. The university also about a tweet from the account criticizing the university’s police the tweet listed the public contact information of the university’s president and police chief.
• The University of New Hampshire blocked a Twitter account belonging to “,” a Libertarian Party presidential candidate, and blocked the Twitter account of Breitbart News—and other —after the outlet wrote about professors who called for the expulsion of counter-protesters who dressed as Harambe and Richard Nixon.^[83] Other accounts appear to have been blocked the university for employing Seth Abramson, whose Twitter criticisms of President Trump have gone viral.
• The University of Montana blocked a Twitter account,, which was being used to follow “the #UMRape crisis that has been plaguing” the university.
• Mississippi State University blocks “.”
• A number of universities block accounts supporting the “Occupy” movement. For example, Ball State University blocks “,” Ivy Tech Community College in Indiana blocks “,” the University of Maryland-College Park blocks “,” and the University of North Carolina blocks “” on Facebook.
• Idaho State University a Facebook page critical of a university employee whose photographs of big game hunting in South Africa went viral.^[84]
• The University of Mississippi, which blocks mentions of its former “Colonel Reb” mascot on its Facebook page, also a operated by an organization (which also has a registered student chapter at the university) that advocates for the mascot’s return.^[85]
• The University of Utah blocks animal rights activists, including , former University of Utah student , and “.” Activists, including PETA and Beckham, have criticized research conducted at the university’s labs.^[86]
• The University of Alaska Anchorage an “Alaskans4Trump” account that had disparaged activist Shaun King in response to a tweet announcing that King would be speaking at the university.^[87]

The reason users have been blocked is not always clear from the records produced to ��֭����. (The University of North Dakota, for example, blocks a Facebook group entitled “Dog Enthusiasts of North Dakota, no posers, only real dog lovers.”) Most institutions block only a handful of users, blocking an average of 76 Facebook users and 23.5 Twitter users. Others wield the tool more liberally. Indiana University blocks the most Facebook users (1488), trailed closely by the University of Washington (1162), with the University of California, Los Angeles in a distant third (691). When it comes to blocked Twitter users, Portland State University blocks the most users (883), more than double that of Central Connecticut State University (307) and Arizona State University (253).

The extent of manual removal of posts and comments is unknown.

FIRE’s survey did not reveal information about the extent to which institutions utilize their ability to manually hide individual Facebook posts or comments after they’ve been posted. However, if institutions are utilizing tools to prevent content from being posted, it is certain that they are also removing content after it has been posted. Recall, for example, that at Ohio’s Wright State University, a public records request revealed that administrators hid hundreds of posts concerning an ongoing faculty union strike.^[88] Unfortunately, this type of content removal is difficult to detect through broad public records requests.

How private and government actors can mitigate the risk of censoring speech on social media

Government actors and social media companies navigating the application of the First Amendment to the digital realm can both take steps to avoid infringing on fundamental civil liberties.

Public actors and private colleges must avoid censoring speech on social media.

For administrators and staff at public universities and colleges—as well as at private institutions that promise freedom of expression to students and faculty—there are a variety of ways to reduce the risk of censorship.

For example:

• Don’t filter. Turn off Facebook’s filters.
• Don’t block users. Don’t block users based on their viewpoint. If they repeatedly violate a policy, document it and give them an opportunity to contest any limits on their access.
• Craft clear, publicly available policies. Be clear about the purpose of your social media pages. Is your page open for comments only from students or faculty, or is it open to the public? If you place limits on the content of speech, make sure they’re published, reasonable, narrow, and objective. A vague policy that leaves your staff with a wide range of discretion is a recipe for trouble. Some universities have begun to formulate new policies.^[89]
• Consistently enforce policies once in place. A failure to enforce policies consistently may lead to First Amendment liability even if the policy itself is clear, objective, and reasonable. Selective enforcement will lead to the appearance of bias, and may demonstrate unlawful viewpoint discrimination.^[90] Further, at least one court has determined that, because policies were inconsistently enforced, a Facebook page was a designated public forum, not a limited public forum, meaning that it was subject to a stricter First Amendment analysis.^[91]
• Train staff on First Amendment principles. Make sure your staff members tasked with regulating and responding to online speech know what they can and can’t do, and make sure they ask for help when they’re not sure.^[92]
• Keep records. If you block a user for posting the same comment repeatedly, or for using unprotected speech, keep a record of it. This will facilitate supervision and correction if a user is blocked for an improper reason.

Social media companies can discourage government actors from abusing tools to censor online speech.

For example:

• Distinguish between government and non-government accounts. If a user identifies themselves as a government actor, make that information public, and require government officials to designate themselves as such in order to create verified pages or accounts.
• Require government actors to establish public-facing policies. Restrict government actors’ moderation tools unless they have created a public policy concerning user content.
• Keep filters “off” by default, and only permit their use if the government actor has established a public-facing content policy.
• Keep filters “off” by default, and only permit their use if the government actor has established a public-facing content policy.
• Alert users when their comment has been filtered on a government actor’s page. If a user does not know that their speech has been hidden, they cannot remedy that censorship.
• Make public the words contained on the profanity filters. Government actors should not be able to make use of a list of forbidden words hidden from the public.
• Provide tools and reminders for government accounts. If an account is designated as belonging to a government actor, provide warnings alongside filtering or blocking tools that their use may have ramifications under the First Amendment. If a government actor blocks a user, require them to set forth their reason for doing so.

Conclusion

Freedom of expression is often one of the few tools available to people to defend themselves. As more of our public discourse takes place online, and not on the sidewalk or in the pages of newspapers, it is critical that public and private institutions alike vigorously defend expressive rights, and remain vigilant to the impact censorship will have on individuals, groups, and communities.

FIRE stands ready to assist students, student organizations, faculty members, and faculty organizations in defending their rights—online or off. If you believe your rights may be jeopardized, contact FIREat /alarm.

Footnotes

^[1] Mark Zuckerberg, Standing for Voice and Free Expression, Georgetown University (Oct. 17, 2019) (transcript available at ).
^[2] This limitation has been upheld by the Supreme Court. , 569 U.S. 221, 224 (2013).
^[3] Facebook, Moderate Your Facebook Page, Dec. 7, 2015, ; see also Facebook, How can I proactively moderate content published by visitors on my Page?, (last visited Apr. 15, 2020).
^[4] Id.
^[5] Id.
^[6] Facebook, How do I ban or unban someone from my Page?, (last visited Apr. 15, 2020).
^[7] Twitter, How to control your Twitter experience, https://help.twitter.com/en/safety-and-security/control-your-twitter-experience (last visited Apr. 15, 2020).
^[8] This function does not infringe on a user’s First Amendment rights: One has a right to speak at or about the government, but the government is not obligated to pay attention to that speech. Accordingly, ��֭����’s survey did not ask about muted users or content.
^[9] , 137 S. Ct. 1730, 1735 (2017); see also , 928 F.3d 226, 237 (2d Cir. 2019) (“Knight”) (“[S]ocial media is entitled to the same First Amendment protections as other forms of media.”).
^[10] To be sure, the First Amendment does not restrict the use of Facebook or Twitter by private actors, such as businesses or private colleges. Nor do Facebook or Twitter themselves violate the First Amendment by providing these tools.
^[11] See, e.g., ., 921 F. 3d 440, 447–49 (5th Cir. 2019) (assuming a sheriff’s Facebook page is a public forum and holding that a policy of deleting “inappropriate” comments was viewpoint discriminatory); , 912 F.3d 666, 681–87 (4th Cir. 2019) (county official’s Facebook page was a public forum); , No. 18-3485, 2018 U.S. App. LEXIS 36090, at *9–14 (6th Cir. 2018) (unpublished opinion reversing sua sponte dismissal of complaint alleging viewpoint discrimination on city’s official Facebook page); Report and Recommendations, , No. A-19-cv-0198-LY-SH, *11–12 (W.D. Tex. Feb. 5, 2020), ECF No. 36 (recommending denial of motion to dismiss First Amendment claim); , No. 4:19-cv-17, at *43–44 (W.D. Ark. Sept. 30, 2019), ECF No. 27 (plaintiffs had “fair chance” of showing state legislator’s Twitter and Facebook blocks were viewpoint discrimination); , No. 17-cv-2215-W (JLB), U.S. Dist. LEXIS 167247, at *14–21 (S.D. Cal. Sept. 26, 2019), ECF No. 42 (denying motion for summary judgment and finding that the interactive portions of school board members’ Facebook and Twitter accounts were designated, not limited, public forums); Order of Dismissal, , No. 19-cv-01367 (D. Colo. Sept. 10, 2019), ECF No. 27 (state legislator agrees to unblock Facebook and Twitter critic, pay $25,000); Findings of Fact and Conclusions of Law, , No. 2:18-cv-4129-BCW, 2019 U.S. Dist. LEXIS 138881, at *9–15 (W.D. Mo. Aug. 16, 2019), ECF No. 55 (interactive space created by state representative’s tweets was a designated public forum); , No. 1:19-cv-24, 2019 U.S. Dist. LEXIS 95080, at *13–18 (N.D.W. Va. June 6, 2019), ECF No. 18 (First Amendment challenge survives motion to dismiss where constituent blocked from legislator’s “politician” Facebook page); , No. 2:18-cv-02138, 2019 U.S. Dist. LEXIS 90509, at *4–5 (E. D. Ark. May 30, 2019), ECF No. 17 (interactive portion of police department’s Facebook page was not government speech); , 354 F. Supp. 3d 940 (W.D. Wis. Jan. 18, 2019) (legislators blocking critic on Twitter); , No. 4:18-cv-01547 (S.D. Tex. Sept. 10, 2018), ECF No. 31 (summary order denying motion to dismiss First Amendment claims premised on keyword-based content filters on public university’s Facebook page); , No. 17-cv-00296, 2018 U.S. Dist. LEXIS 146894, *36–43 (D. Me. Aug. 29, 2018) (governor’s Facebook page was limited public forum); , No. 15-cv-5871, 2018 U.S. Dist. LEXIS 105815, *25–46 (S.D.N.Y. June 25, 2018) (where NYPD precinct blocked Twitter user, whether the forum was a “public, designated, or nonpublic forum” was immaterial, as “viewpoint discrimination that results in the intentional, targeted expulsion of individuals . . . is unlawful in any forum,” including nonpublic forums); , No. 3:17-cv-01531, 2018 U.S. Dist. LEXIS 95832 (D. Conn. June 7, 2018) (critic blocked from police Facebook page); , No. 1:16-cv-180, 2017 U.S. Dist. LEXIS 4348, at *10 (E.D. Va. Jan. 10, 2017) (county’s policy and practice of encouraging comments on its Facebook page created a limited public forum).
^[12] , 302 F. Supp. 3d 541, 549 (S.D.N.Y. May 23, 2018) (upheld on appeal by the Second Circuit in Knight).
^[13] See, e.g., , No. 18-15712, 2020 U.S. App. LEXIS 5903, at *7 (9th Cir. Feb. 26, 2020) (no First Amendment claim that YouTube “censored” speech because the free speech clause “prohibits the government—not a private party—from abridging speech.”).
^[14] Aaron H. Caplan, , 46 Willamette L. Rev. 647, 647 (2009-2010).
^[15] , 333 F.3d 1092, 1099 (9th Cir. 2003) (the “quintessential traditional public forums are side-walks, streets, and parks.”).
^[16] See, e.g., , 586 F.3d 427, 432 (6th Cir. 2009) (school board meetings give “citizens a chance to express their views to the board, [but] cannot accommodate the sort of uninhibited, unstructured speech that characterizes a public park.”).
^[17] , 420 U.S. 546, 555 (1975) (government-leased theater “under [the] control” of public officials); , 518 U.S. 727, 749 (1996).
^[18] Government actors have sought to frame ���������ٴǰ���’ Facebook or Twitter comments as government speech, arguing that because the government hasn’t removed those comments, it is effectively publishing the comments as the government’s own speech. Most people, however, would not view a Facebook comment or tweet as government-endorsed simply because it hasn’t been removed.
^[19] United States v. Alvarez, 567 U.S. 709, 709 (2012).
^[20] , 605 F.3d 703, 710 (9th Cir. 2009).
^[21] Rosenberger v. Rector & Visitors of Univ. of Va., 515 U.S. 819, 829–30 (1995). This is true even if the interactive space of the social media page is deemed a non-public forum, as access even to a non-public forum may only be “restricted as long as the restrictions are ‘reasonable and [are] not an effort to suppress expression merely because public officials oppose the speaker’s view.’” Cornelius v. NAACP Legal Def. & Educ. Fund, Inc., 473 U.S. 788, 800 (1985) (quoting Perry Educ. Ass’n v. Perry Local Educators’ Ass’n, 450 U.S. 37, 46 (1983)).
^[22] Good News Club v. Milford Cent. Sch., 533 U.S. 98, 106 (2001) (distinguishing “traditional or open” public forums from “limited” public forums); , No. 17-cv-2215-W (JLB), U.S. Dist. LEXIS 167247, at *14–21 (S.D. Cal. Sept. 26, 2019), ECF No. 42 (interactive portions of school board members’ Facebook and Twitter accounts were designated, not limited, public forums); , 354 F. Supp. 3d 940, 953–55 (W.D. Wis. Jan. 18, 2019) (interactive portions of legislators’ Twitter accounts were designated public forums).
^[23] Perry Educ. Ass’n, 450 U.S. at 45.
^[24] Id.
^[25] , U.S. Dist. LEXIS 167247 at *20–21.
^[26] , 480 F. 3d 891, 908 n.8 (9th Cir. 2006) (A “limited public forum is a sub-category of the designated public forum, where the government opens a nonpublic forum but reserves access to it for only certain groups or categories of speech.”).
^[27] Good News Club, 533 U.S. at 106–07 (quoting, in part, Rosenberger v. Rector & Visitors of Univ. of Va., 515 U.S. 819, 829 (1995)).
^[28] , 699 F.3d 1053, 1062 (9th Cir. 2012).
^[29] See, e.g., , 482 F.Supp.2d 1003, 1009¬¬–11 (S.D. Ohio 2007) (government agency’s bulletin board had “some aspects of a limited public forum” because it was “open to all employees” to “post matters addressing a broad range of topics or meetings,” rendering removal of religious materials unconstitutional).
^[30] Rosenberger v. Rector & Visitors of Univ. of Va., 515 U.S. 819, 830 (1995).
^[31] Id.
^[32] Perry Educ. Ass’n v. Perry Local Educators’ Ass’n, 450 U.S. 37, 46¬–49 (1983).
^[33] , 605 F.3d 703, 710 (9th Cir. 2009).
^[34] Compare , 221 F.3d 834, 844 (6th Cir. 2000) (city’s website listing links to other sites was a non-public forum, not a public forum, because it was not structured to allow “dialogue between users,” but primarily “to convey information to the reader”) with , 531 F.3d 275, 284 (4th Cir. 2008) (school district’s website was not a public forum, but “the issue would, of course, be different” if the website were a “type of ‘chat room’ or ‘bulletin board’ in which private viewers could express opinions or post information”).
^[35] , 302 F. Supp. 3d 541, 572–73 (S.D.N.Y. May 23, 2018).
^[36] , 912 F.3d 666, 687–88 (4th Cir. 2019) (official who blocked constituent because of his critical viewpoint was viewpoint discrimination, which is “prohibited in all forums.”).
^[37] , 921 F. 3d 440, 447 (5th Cir. 2019).
^[38] at 238–39.
^[39] , 247 F.Supp.3d 767, 777 (E.D. Va. 2017).
^[40] See, e.g., , 990 F.2d 586, 591 (11th Cir. 1993) (a restriction vesting unfettered discretion in a government actor “opens the way to arbitrary suppression of particular points of view.”).
^[41] Greg Harold Greubel, Public records reveal Wright State used unconstitutional Facebook page policy to censor pro-union speech during historic faculty strike, Found. for Indiv. Rights in Educ., Jan. 24, 2020, /public-records-reveal-wright-state-used-unconstitutional-facebook-page-policy-to-censor-pro-union-speech-during-historic-faculty-strike; see also, Letter from Greg Harold Greubel, Staff Attorney, Foundation for Individual Rights in Education to Larry Y. Chan, General Counsel, Wright State University, Jan. 23, 2020, available at /fire-letter-to-wright-state-university-january-23-2020.
^[42] Healy v. James, 408 U.S. 169, 180 (1972) (there is “no room for the view that . . . First Amendment protections should apply with less force on college campuses than in the community at large”).
^[43] As discussed in the methodology section of this report and in the appendices, some states or institutions could not be surveyed, failed to respond at all, sought exorbitant fees, or cited exceptions to open records laws, ranging from arguments that they cannot be required to “create” records to claims that revealing Facebook blacklists creates a risk to public safety.
^[44] See, Miami Herald Pub. Co., Div. of Knight Newspapers Inc. v. Tornillo, 418 U.S. 241, 256–57 (1974) (a “[g]overnment-enforced right of access” in the form of a statute compelling newspapers to publish rebuttals “operates as a command in the same sense as” a rule prohibiting speech, violating the First Amendment). This right is fortified by statute shielding providers of an “interactive computer service” from civil liability for removing content they deem “objectionable.” .
^[45] Cohen v. California, 403 U.S. 15, 16 (1971).
^[46] Id. at 26.
^[47] Texas v. Johnson, 491 U.S. 397, 414 (1989).
^[48] Maarten Sap, Dallas Card, Saadia Gabriel, Yejin Choi, and Noah A. Smith, 2019, The Risk of Racial Bias in Hate Speech Detection. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics at 1668–78, available at .
^[49] Lucas Dixon, John Li, Jeffrey Sorensen, Nithum Thain, and Lucy Vasserman. 2018. Measuring and Mitigating Unintended Bias in Text Classification. In Proceedings of AAAI/ACM Conference on Artificial Intelligence, Ethics, and Society at 67, available at .
^[50] Thomas Davidson, Debasmita Bhattacharya, and Ingmar Weber, 2019, Racial Bias in Hate Speech and Abusive Language Detection Datasets, available at .
^[51] Mike Masnick, Masnick’s Impossibility Theorem: Content Moderation At Scale Is Impossible To Do Well, Techdirt, Nov. 20, 2019, .
^[52] Shirin Ghaffary, The algorithms that detect hate speech online are biased against black people, Vox, Aug. 15, 2019, .
^[53] See, e.g., Matal v. Tam, 137 S. Ct. 1744, 1763 (2017) (striking down regulation barring trademark registration for “any mark that is offensive to a substantial percentage of the members of any group” because it amounted to viewpoint discrimination, and observing that “[g]iving offense is a viewpoint”).
^[54] Grayned v. City of Rockford, 408 U.S. 104, 108 (1972).
^[55] Neb. Press Ass’n v. Stuart, 427 U.S. 539, 559 (1976).
^[56] Near v. Minnesota, 283 U.S. 697, 713 (1931).
^[57] Shuttlesworth v. Birmingham, 394 U.S. 147, 151 (1969).
^[58] See, e.g., , No. 3:18-cv-1017 (PAD), 2018 U.S. Dist. LEXIS 208014, at *21–24 (D.P.R. Dec. 10, 2018) (surveying the procedural and substantive strands of the doctrine of prior restraint); see also, , 493 F.3d 421, 431 (4th Cir. 2007) (“A prior restraint on speech that imposes no time limitations on the decision-making process plainly fails” First Amendment scrutiny when the prior restraint is based on content.).
^[59] In the Matter of a Citizen’s Complaint Against Pacifica Foundation Station WBAI (FM), New York, New York, 56 F.C.C.2d 94, 109, Declaratory Order (Feb. 21, 1975) (Transcript of George Carlin’s “Filthy Words” monologue, prepared by the Federal Communications Commission), archived at .
^[60] See, e.g., Univ. of N.C. at Chapel Hill, Recognizing Carolina’s black pioneers, Facebook (Feb. 23, 2018), .
^[61] After a letter from ��֭����, UNC removed some of the terms on its customized list, but it is not clear what terms remain. Letter from Kara E. Simmons, Associate Vice Chancellor and Senior Univ. Counsel, Univ. N.C., to Adam Steinbaugh, Director, Individual Rights Defense Program, Found. for Indiv. Rights in Educ. (Sept. 13, 2019) (on file with author).
^[62] Univ. of N.C. at Chapel Hill, Chancellor Folt announces resignation. . ., Facebook (Jan. 15, 2019), .
^[63] Univ. of N.C. at Chapel Hill, Carolina community gathers in remembrance of 9/11, Facebook (Sept. 11, 2018), .
^[64] See, e.g., Oklahoma State Univ., It’s Election Day, Cowboys! Get out and VOTE!, Facebook (Nov. 8, 2016), .
^[65] Good News Club v. Milford Cent. Sch., 533 U.S. 98, 106–07 (2001) (quoting, in part, Rosenberger v. Rector & Visitors of Univ. of Va., 515 U.S. 819, 829 (1995)).
^[66] Jay Blanton, Dining Partnership Will Transform Vital Service to UK Campus, Univ. of Ky., June 13, 2014, .
^[67] Adam Schwartz, EFF Lawsuit Ends Censorship Against PETA on Public University’s Facebook Page, Electronic Frontier Found., Feb. 4, 2020, ; see also Adam Steinbaugh, Texas A&M: Our secret list of naughty words you can’t say on our Facebook page doesn’t offend the First Amendment, Found. for Indiv. Rights in Educ., Aug. 23, 2018, /texas-am-our-secret-list-of-naughty-words-you-cant-say-on-our-facebook-page-doesnt-offend-the-first-amendment.
^[68] Press Release, Secret Student Video Exposes Dissection of Pregnant Cat at Santa Monica College, People for Ethical Treatment of Animals, Sept. 7, 2016, .
^[69] Neel Ahuja, Distorting the study of 9/11 at UNC, News & Observer, Sept. 19, 2015, .
^[70] Jane Stancill, UNC course on 9/11 criticized in conservative publications, Charlotte Observer, Sept. 1, 2015, .
^[71] Aviva Shen, University Of Arizona Student Tells Women: ‘You Deserve Rape’, ThinkProgress, Apr. 25, 2013, .
^[72] Manie Robinson, Clemson extends Nike partnership with 10-year, $58 million deal, Greenville News, Aug. 3, 2018, .
^[73] Georgie Silvarole, Clemson University student senators sit for Pledge of Allegiance, WCNC, Sept. 28, 2017, .
^[74] Matt Connolly, Clemson ending this football gameday tradition for 2018, The State, July 28, 2018, .
^[75] E-mail from Jermaine D. Johnson, Assistant General Counsel, Clemson Univ., to Adam Steinbaugh, Director, Individual Rights Defense Program, Found. for Indiv. Rights in Educ. (Dec. 21, 2018, 10:12 PM), available at .
^[76] Invisible Children, Kony 2012, (last visited Apr. 15, 2020).
^[77] Robbie Brown, Ole Miss Shelves Mascot Fraught With Baggage, N.Y. Times, Sept. 19, 2019, .
^[78] Extending the middle finger is expression protected by the First Amendment. , 704 F.3d 105, 111 (2d Cir. 2013).
^[79] Jason Kirk, Some moron spray-painted ‘Fail State,’ ‘Go Rebs,’ and more on Mississippi State’s campus, SBNation, Dec. 17, 2016, .
^[80] Jesse Wright, With Hopes to Curb Couch Burning, Morgantown Passes Outdoor Furniture Ban, W.VA. Pub. Broad., Apr. 7, 2015, .
^[81] America’s Largest Private Companies, Forbes, (last visited Apr. 15, 2020).
^[82] Press Release, Koch Industries Gift of $1M Establishes Business, Engineering Scholarship Funds, Univ. of Kan. (Sept. 8, 2016), . FIREalso receives funding from the Charles Koch Foundation. FIREdoes not accept money from donors who seek to alter our mission or interfere with our case work.
^[83] Katherine Rodriguez, University of New Hampshire Professors Call for Trump Supporters to Be Expelled, Breitbart, Nov. 21, 2016, .
^[84] Michael H. O’Donnell, ISU accountant on safari becomes social media target, Idaho St. J., Aug. 5, 2015, .
^[85] Colonel Reb Found., About Us, (last visited Apr. 15, 2020); see also Univ. of Miss., Colonel Reb Found., (recognized student organization webpage) (last visited Apr. 15, 2020).
^[86] Courtney Tanner, University of Utah can’t charge PETA a $5K ‘prepayment’ for records on lab animal deaths, committee rules, Salt Lake Trib.., Nov. 14, 2018, .
^[87] @Alaskans4Trump, Twitter (Feb. 7, 2017, 1:30 AM), .
^[88] Greg Harold Greubel, Public records reveal Wright State used unconstitutional Facebook page policy to censor pro-union speech during historic faculty strike, Found. for Indiv. Rights in Educ., Jan. 24, 2020, /public-records-reveal-wright-state-used-unconstitutional-facebook-page-policy-to-censor-pro-union-speech-during-historic-faculty-strike.
^[89] See, e.g., and the .
^[90] Thomas v. Chi. Park Dist., 534 U.S. 316, 325 (2002) (granting “waivers to favored speakers” or “denying them to disfavored speakers” would “of course be unconstitutional”); see, e.g., , 360 F.Supp.3d 885, 907 (S.D. Iowa 2019) (disparate enforcement of nondiscrimination policy violated the First Amendment).
^[91] , No. 17-cv-2215-W (JLB), U.S. Dist. LEXIS 167247, at *20 (S.D. Cal. Sept. 26, 2019), ECF No. 42.
^[92] To its credit, Wayne State University started training its social media staff after receiving a records request from ��֭����.

Appendices

Appendix A

Surveyed institutions and response data

You can review the survey results, including links to each institution's records, here: .

Institutions in blue provided adequate responses. Copies of the records are hyperlinked.

Institutions marked in red failed to produce sufficient documents for the reasons specified in the accompanying footnote or hyperlinked document.

Appendix B

Exceptions to the survey methodology

The survey sought to cover two sets of institutions within each state, plus the District of Columbia: the three four-year institutions with the highest enrollment and the two two-year institutions with the highest enrollment. This would amount to 255 institutions, but not every state has qualifying institutions, some states’ public records laws do not extend to those institutions, and some institutions cannot be surveyed due to a legal conflict. What follows is a list of states or institutions falling under an exception or conflict:

• Arkansas: Public records law limited to citizens of Arkansas. (-5 institutions).
• California: East Los Angeles College could not be contacted due to litigation conflict. Replaced by American River College.
• Delaware:
  • Only two four-year institutions, and only one two-year institution qualified. (-2 institutions).
  • The qualifying four-year institutions (Delaware State University and the University of Delaware) are . (-2 institutions).
• Iowa: Iowa State University could not be contacted due to litigation conflict. Cannot be replaced by another four-year institution because there are only three public four-year institutions in Iowa. (-1 institution).
• Louisiana: Louisiana State University could not be contacted due to litigation conflict. Replaced by Louisiana Tech.
• New Jersey: Rutgers University could not be contacted due to possibility of litigation conflict. Replaced with Rowan University.
• Pennsylvania: The Right to Know law largely exempts four-year institutions, but does not exempt two-year institutions. (-3 institutions).
• Rhode Island:
  • The Naval War College was not surveyed because its relationship with the armed services may change the First Amendment calculus. It was not replaced because there are only three, four-year institutions in the state. (-1 institution).
  • Rhode Island only has one two-year institution. (-1 institution).
• South Carolina: The University of South Carolina could not be surveyed due to litigation conflict. Replaced by Coastal Carolina University.
• Tennessee: FOIA limited to citizens of Tennessee. (-5 institutions).
• Virginia: FOIA limited to citizens of Virginia. (-5 institutions).
• Washington, D.C.:
  • There’s only one public, four-year institution in Washington, D.C., that is not affiliated with the military. (-2 institutions).
  • There are no public two-year institutions. (-2 institutions).
• Wyoming: There is only one public, four-year institution. (-2 institutions).

Appendix C

Example public records request

The following is a lightly edited version of the request sent to Michigan State University:

This is a request for records pursuant to the Michigan Freedom of Information Act (Mich. Comp. Laws Ann. §§ 15.231 et seq.). If you are not the records custodian for MSU, please identify the correct person to contact.

This request seeks records relating to restrictions and settings concerning the official Facebook and Twitter accounts for MSU, and should be directed to the person responsible for operating those accounts.

RECORDS REQUESTED

I request the following records:

1. A copy of the settings for the Facebook page maintained by Michigan State University (available at https://www.facebook.com/spartans.msu). This list is accessible by (A) logging into the Facebook page as an administrator, and then (B) clicking “Settings” at the top of the official page. The URL should look like: https://www.facebook.com/spartans.msu/settings/?tab=settings.
2. A copy of the list of people or pages banned from the Facebook page referenced above. This list is accessible by: (A) logging into the Facebook page as an administrator, (B) clicking “Settings” at the top of the official page, (C) clicking “People and Other Pages” in the left column, and (D) selecting "Banned People and Pages" from the drop-down menu. The final URL should look like: https://www.facebook.com/spartans.msu/settings/?tab=people_and_other_pages.
3. A list of the "blocked accounts" by the Twitter account maintained by Michigan State University (available at https://twitter.com/michiganstateu). This list is accessible by navigating to this URL while logged into the account: https://twitter.com/settings/blocked.

Fee waiver request: This request concerns a matter of public interest. The social media restrictions imposed by public institutions and officials — including the president, governors, and public universities — have been challenged on First Amendment grounds. This survey seeks to explore the extent to which public colleges and universities have similar restrictions.

The public interest would be well-served by granting a fee waiver. The request is not being sought for a commercial purpose, but is instead sought by a nonprofit organization to provide the public with information concerning the conduct of government actors as that conduct pertains to civil liberties in higher education.

If a fee waiver is not granted, please apprise me if the estimated costs will exceed $10.

Request for expedited processing: Completion of this survey depends on the institution with the slowest response time. We request that Michigan State University produce responsive records on an expedited basis. As you may be aware, a public body has five business days to respond. (Mich. Comp. Laws. Ann § 15.235(2)).

Appeal information: If you deny any portion, or all, of this request, please provide me with a written explanation of the reason(s) for your denial, including a citation to each specific statutory exemption you believe justifies the refusal to release the information and notify me of the appeal procedures available to me under the law. If you conclude that portions of the records that I request are exempt from disclosure, please release the remainder of such records for inspection and copying, redacting only the portion or portions that you claim are exempt.

Appendix D

Customized blocked words

The surveyed institutions blocked the following unique words using Facebook’s customizable blacklist. Duplicates were removed.